海洋学研究 ›› 2021, Vol. 39 ›› Issue (3): 12-20.DOI: 10.3969/j.issn.1001-909X.2021.03.002

• • 上一篇    下一篇

基于海底DEM的洋中脊火山锥自动识别方法研究

党牛1,2,3，余星*2,3，韩喜球2,3，陈安清1   

  1. 1.成都理工大学 沉积地质研究院, 四川 成都 610059;
    2.自然资源部 海底科学重点实验室, 浙江 杭州 310012;
    3.自然资源部 第二海洋研究所，浙江 杭州 310012
  • 出版日期:2021-09-15 发布日期:2021-09-15
  • 基金资助:
    国家自然科学基金项目 (41872242,42172231)；中央级公益性科研院所基本科研业务费专项资金项目(JT2001，JG2001)；大洋“十三五”资源环境专项(DY135-S2-1-02)

Automatic recognition of volcanic cones at mid-ocean ridges based on the seabed DEM data

DANG Niu1,2,3, YU Xing*2,3, HAN Xiqiu2,3, CHEN Anqing1   

  1. 1.Institute of Sedimentary Geology, Chengdu University of Technology, Chengdu 610059, China;
    2.Key Laboratory of Submarine Geosciences, Ministry of Natural Resources, Hangzhou 310012, China;
    3.Second Institute of Oceanography, Ministry of Natural Resources, Hangzhou 310012, China
  • Online:2021-09-15 Published:2021-09-15

PDF

466

摘要： 洋中脊是板块扩张和洋壳增生的主要区域，除了发育沿洋脊走向的裂隙式喷发火山脊，还分布有众多零星的中心式喷发火山锥。这些火山锥的形态和分布对了解洋中脊构造和岩浆活动具有重要的指示意义。基于海底多波束地形数据，采用人工或机器解译方法可以识别这些火山锥。本文利用大洋24航次船载多波束测深获得的卡尔斯伯格脊DEM(数字高程模型)数据，以非监督分类为手段，开展洋中脊火山锥的自动提取方法研究。基于原始DEM计算地形坡度、地表粗糙度、正负地形等衍生参数，进行特征变换，提取火山锥的特征信息。使用ISO(迭代自组织)聚类方法对特征变换后的图像进行聚类分析，并利用景观形状指数进行几何筛选，完成火山锥的自动提取。所使用的海底火山锥自动识别方法，正确率达0.8，重叠率约0.7，识别效果较好，效率高，能够在海底大范围的火山锥解译中发挥重要作用。

关键词: 火山锥, 洋中脊, 非监督分类, 特征变换, DEM, 自动识别

Abstract: The mid-ocean ridge is where the plates spread and the new oceanic crust forms. There are often well-developed central eruptive volcanic cones in addition to fissure eruptive lavas parallel to the ridge axis. These volcanic cones are of great significance for understanding the local and regional magmatism and tectonic activities. These volcanic cones can be identified using manual or machine interpretation methods based on seabed multi-beam bathymetric data. In this study the automatic extraction method of volcanic cones near the mid-ocean ridge was tried by means of unsupervised classification using the DEM data from Carlsberg Ridge obtained from the Chinese DY24 Cruise. The slope, surface roughness, positive and negative topography and other derivative parameters were calculated based on the original DEM data. The morphological characteristics and spatial features of volcanic cones were enhanced by feature transformation. The ISO clustering unsupervised classification method was selected to cluster and analyze the images after feature transformation. Then the method of landscape shape index in landscape ecology was introduced to complete the automatic extraction of volcanic cones. The accuracy of automatic recognition can reach around 0.8 with overlapping rate of ~0.7 comparing with manual extraction. Thus, the automatic extraction of submarine volcanic cones by means of unsupervised classification is robust and efficient, which can be of great help to large-scale data process and interpretation.

Key words: volcanic cone, mid-ocean ridge, unsupervised classification, feature transformation, DEM, automatic recognition

中图分类号: